linux-mmap-antirecursion-fix-20090512

LICENSE IPL10
FIXES 124627

fix the mmap anti-recursion protection to set and mask CPageWrite appropriately
and with the correct lock protection. this leaves us with an issue to handle
when the mmap'd file is larger (possibly considerably larger) than the
cache

diff --git a/src/afs/LINUX/osi_vm.c b/src/afs/LINUX/osi_vm.c
index c96babd..55a3737 100644 (file)
--- a/src/afs/LINUX/osi_vm.c
+++ b/src/afs/LINUX/osi_vm.c
@@ -102,9 +102,7 @@ osi_VM_StoreAllSegments(struct vcache *avc)
 {
     struct inode *ip = AFSTOV(avc);
 
-    if (!avc->f.states & CPageWrite)
-       avc->f.states |= CPageWrite;
-    else 
+    if (avc->f.states & CPageWrite)
        return; /* someone already writing */
 
 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,5)
@@ -120,7 +118,6 @@ osi_VM_StoreAllSegments(struct vcache *avc)
     AFS_GLOCK();
     ObtainWriteLock(&avc->lock, 121);
 #endif
-    avc->f.states &= ~CPageWrite;
 }
 
 /* Purge VM for a file when its callback is revoked.

diff --git a/src/afs/LINUX/osi_vnodeops.c b/src/afs/LINUX/osi_vnodeops.c
index 67e7bad..39bd081 100644 (file)
--- a/src/afs/LINUX/osi_vnodeops.c
+++ b/src/afs/LINUX/osi_vnodeops.c
@@ -1188,7 +1188,7 @@ afs_linux_lookup(struct inode *dip, struct dentry *dp)
     if (code == ENOENT)
        return ERR_PTR(0);
 #endif
-    else if ((code >= 0) && (code <= MAX_ERRNO))
+    else if ((code > 0) && (code <= MAX_ERRNO))
         return ERR_PTR(-code);
     else 
        return ERR_PTR(-EIO);
@@ -1895,9 +1895,9 @@ afs_linux_writepage_sync(struct inode *ip, struct page *pp,
               ICL_TYPE_POINTER, pp, ICL_TYPE_INT32, page_count(pp),
               ICL_TYPE_INT32, 99999);
 
-    ObtainReadLock(&vcp->lock);
+    ObtainWriteLock(&vcp->lock, 532);
     if (vcp->f.states & CPageWrite) {
-       ReleaseReadLock(&vcp->lock);
+       ReleaseWriteLock(&vcp->lock);
        AFS_GUNLOCK();
        maybe_unlock_kernel();
        crfree(credp);
@@ -1913,7 +1913,8 @@ afs_linux_writepage_sync(struct inode *ip, struct page *pp,
        return(0); 
 #endif
     }
-    ReleaseReadLock(&vcp->lock);
+    vcp->states |= CPageWrite;
+    ReleaseWriteLock(&vcp->lock);
 
     setup_uio(&tuio, &iovec, buffer, base, count, UIO_WRITE, AFS_UIOSYS);
 
@@ -1922,16 +1923,18 @@ afs_linux_writepage_sync(struct inode *ip, struct page *pp,
     i_size_write(ip, vcp->f.m.Length);
     ip->i_blocks = ((vcp->f.m.Length + 1023) >> 10) << 1;
 
+    ObtainWriteLock(&vcp->lock, 533);
     if (!code) {
        struct vrequest treq;
 
-       ObtainWriteLock(&vcp->lock, 533);
        if (!afs_InitReq(&treq, credp))
            code = afs_DoPartialWrite(vcp, &treq);
-       ReleaseWriteLock(&vcp->lock);
     }
     code = code ? -code : count - tuio.uio_resid;
 
+    vcp->states &= ~CPageWrite;
+    ReleaseWriteLock(&vcp->lock);
+
     afs_Trace4(afs_iclSetp, CM_TRACE_UPDATEPAGE, ICL_TYPE_POINTER, vcp,
               ICL_TYPE_POINTER, pp, ICL_TYPE_INT32, page_count(pp),
               ICL_TYPE_INT32, code);
@@ -1986,7 +1989,12 @@ afs_linux_writepage(struct page *pp)
   do_it:
     status = afs_linux_writepage_sync(inode, pp, 0, offset);
     SetPageUptodate(pp);
-    UnlockPage(pp);
+#if defined(WRITEPAGE_ACTIVATE)
+    if ( status != WRITEPAGE_ACTIVATE )
+#else
+    if ( status != AOP_WRITEPAGE_ACTIVATE )
+#endif
+       UnlockPage(pp);
     if (status == offset)
        return 0;
     else